Ancient Rome classified the slave as a res loquens: although a natural person, the servant did not have autonomous legal capacity. Subject to the dominica potestas of the master – exercised over him as over an object – the slave could administer a peculium, but any contract or tort fell under the dominium of the dominus. Instruments such as manumissio (formal liberation) and actio noxalis (master's responsibility for the servant's actions) formally regulated status, but never attributed any juridical subjectivity to the servant: in essence, he was a human being treated as a thing.

Similarly, Regulation (EU) 2024/1689 (AI Act) defines artificial intelligence systems as automated entities that produce outputs – content, predictions, decisions, recommendations – capable of influencing physical or virtual environments.

However, similarly to the Roman slave, even intelligent androids – increasingly close to passing from experimental laboratories to markets and the homes of human beings – remain juridically res: technological tools or products subject to security obligations rather than autonomous subjects of law. While it is true that at European level there has even been speculation of elevating robots and AI to "electronic people", the current regulatory framework remains profoundly anthropocentric. In fact, as modern doctrine points out, Western legal systems still consider man as the "fulcrum of society", relegating non-human entities to objects of protection and control (e.g. through rigorous certifications and supplier responsibilities). The current AI Act continues - in a certain sense - the path started by Roman law: it precisely defines obligations and risks for those who produce or use advanced AI, but does not affect the paradigm that conceives the intelligent android as an asset to be governed, rather than as a subject with rights.

This historical-legal parallelism illuminates shared profiles: both systems – ancient and modern – reflect an anthropocentric legal approach, according to which only the human being is the effective holder of rights, while the other "speakers" are seen as res with rights mediated by the dominus/deployers.

The comparative and critical objective of the essay is precisely to bring out these analogies: to compare ancient and modern norms and institutions (res loquens vs AI systems, dominica potestas vs AI governance obligations, peculium vs algorithmic resources, manumissio vs disabling mechanisms, etc.) to reflect on how the subjectivity of non-human intelligent entities (or not considered human) has been legally managed over time and what are the regulatory perspectives,  philosophical and social aspects that await us.